Evaluating Surgical Performance in Real Time Using Data Mining

Zhou, Y., Ioannou, I., Bailey, J., Kennedy, G. and OLeary, S.

Virtual reality simulators are becoming increasingly popular as adjuncts to traditional surgical training methods, but most simulators do not have the ability to evaluate performance on-the-fly and provide advice to trainees as they practice. Timely feedback on performance is a critical component of surgical training, therefore the ability to provide such evaluation is necessary if simulation is to be effective as a platform for self-guided learning. We propose an evaluation framework to automatically assess performance within a temporal bone simulator in real time. This evaluation framework uses data mining techniques to assess performance at different granularities. Drilling technique is analysed to deliver detailed short-term evaluation, while hidden markov models are used to evaluate the completion of small surgical subtasks and provide medium-term assessment. Finally, an analysis of drilled bone shape is used to evaluate performance at the completion of each stage of a surgical procedure. We demonstrate the effectiveness of the proposed methods by validating them on an existing simulation dataset.

Cite as: Zhou, Y., Ioannou, I., Bailey, J., Kennedy, G. and OLeary, S. (2013). Evaluating Surgical Performance in Real Time Using Data Mining. In Proc. Eleventh Australasian Data Mining Conference (AusDM13) Canberra, Australia. CRPIT, 146. Christen, P., Kennedy, P., Liu, L., Ong, K.L., Stranieri, A. and Zhao, Y. Eds., ACS. 25-34

(from crpit.com) (local if available)